[Correction of self-reported prevalence in epidemiological studies with large samples]

Area of Science:

• Epidemiology
• Biostatistics
• Public Health Policy

Background:

• Disease prevalence rates are crucial for public policy.
• Self-reported data is a common, cost-effective method for prevalence measurement.
• Self-reporting can introduce significant bias into prevalence estimates.

Purpose of the Study:

• To review existing methods for adjusting self-reported prevalence data.
• To address computational challenges of current methods with large sample sizes.
• To propose an alternative, computationally feasible solution for large-scale prevalence estimation.

Main Methods:

• Classification of existing methods into algebraic and Bayesian approaches.
• Identification of computational limitations of Bayesian methods for large datasets.
• Development and validation of an empirical approximation strategy for large samples, involving sample reduction while maintaining patient proportions.

Main Results:

• Algebraic methods have limitations in applicability.
• Bayesian methods face computational hurdles with large sample sizes.
• The proposed empirical method demonstrated convergence with true prevalence values, correcting a 5% self-reported prevalence to 0.17% (95%CI: 0.10-0.24) with given sensitivity and specificity.

Conclusions:

• Existing methods for adjusting self-reported prevalence have limitations, especially for large datasets.
• The novel empirical strategy enables accurate prevalence estimation in large populations.
• This approach provides more reliable estimates closer to true values without exhaustive individual measurement.